`+decorate` meta to enable auto-decoration of all objects

[yegor256]

Now, I do this:

[] > foo
  join.
    QQ.txt.text ","
    split.
      QQ.txt.text
        "hello jeff"

Looks rather verbose, since I have to decorate string with QQ.txt.text every time I use it. How about this sugar instead:

+decorate QQ.string ~> QQ.txt.text

[] > foo
  join.
    ","
    split.
      "hello jeff"

Or simply this:

+decorate string ~> QQ.txt.text

This meta means: "every object of the string forma must be automatically decorated with QQ.txt.text"

[yegor256]

@maxonfjvipon WDYT?

[maxonfjvipon]

@yegor256 I like the idea. The only controversial thing is ~> sign. I believe it should be simpler, because tilda is not the most convenient key to press on keyboard

[yegor256]

@maxonfjvipon we can do > instead, but this symbol is already used for object naming. May lead to confusion. How about @> instead?

+decorate string @> QQ.txt.text @> foo

[maxonfjvipon]

@yegor256 where should we put the transformation(s) which would do this decorating?

[yegor256]

@maxonfjvipon most probably, right in the parser. Objects in the XMIR that I get from the parser should already be decorated. It's syntax-level decoration, not runtime.

[maxonfjvipon]

@yegor256 maybe we don't need this @> sign? It just add complexity to the meta, so we need to verify it before appending decorators.
What if we just do: +decoratee string QQ.txt.text foo, which means the same: string is wrapped by text, and all of this is wrapped by foo
WDYT?

[yegor256]

@maxonfjvipon I see the logic, but look at the syntax of Ruby, for example:

class Circle < Figure
  # body
end

They also use a special symbol <, while technically they don't need it. It could have been:

class Circle Figure
  # body
end

I'm not sure though. Maybe you are right :)

[maxonfjvipon]

@yegor256 let's implement it without extra syntax and when it works, if it's not convenient or confusing - extend the meta with some extra sign